Anatomical and functional assessment of CAD using machine learning

Abstract

Anatomical and functional assessment of coronary artery disease (CAD) using machine learning and computational modeling techniques deploying methodologies for non-invasive Fractional Flow Reserve (FFR) quantification based on angiographically derived anatomy and hemodynamics data, relying on machine learning algorithms for image segmentation and flow assessment, and relying on accurate physics-based computational fluid dynamics (CFD) simulation for computation of the FFR.

What is claimed: 
     
       1. A computer-implemented method for assessing coronary artery disease, the method comprising:
 (a) receiving, by one or more processors, angiography image data of a vessel inspection region for a subject, wherein the angiography image data comprises angiography images captured over a sampling time period; 
 (b) applying, by the one or more processors, the angiography image data to a vessel segmentation machine learning model and generating, using the vessel segmentation machine learning model, two-dimensional (2D) segmented vessel images for the vessel inspection region; 
 (c) by the one or more processors, generating from the 2D segmented vessel images a three-dimensional (3D) segmented vessel tree geometric model of vessels within the vessel inspection region; 
 (d) applying, by the one or more processors, the angiography images to a fluid dynamics machine learning model and assimilating flow data for the vessel inspection region over the sampling time period; 
 (e) applying, by the one or more processors, the 3D segmented vessel tree geometric model and the assimilated flow data to a computational fluid dynamics model; and 
 (f) determining, by the one or more processors, a state of vessel occlusion for one or more of the vessels within the vessel inspection region. 
 
     
     
       2. The computer-implemented method of  claim 1 , further comprising: determining, by the one or more processors, a state of microvascular disease for the one or more vessels within the vessel inspection region by performing (a)-(e) at least two different hemodynamic states. 
     
     
       3. The computer-implemented method of  claim 1 , wherein the vessel segmentation machine learning model is a convolutional neural network. 
     
     
       4. The computer-implemented method of  claim 1 , further comprising:
 applying, by the one or more processors, to the received angiography image data at least one of a de-noising process, a linear filtering process, an image size normalization process, and a pixel intensity normalization process to produce filtered angiography image data. 
 
     
     
       5. The computer-implemented method of  claim 4 , further comprising:
 feeding the filtered angiography image data to an angiography processing network (APN), trained to address in the angiography image data low contrast, presence of catheters, and/or presence of overlapping bonny bony structures. 
 
     
     
       6. The computer-implemented method of  claim 5 , further comprising:
 feeding an output of the APN to a semantic image segmentation to produce automatic binary 2D segmented vessel images. 
 
     
     
       7. The computer-implemented method of  claim 1 , further comprising:
 applying, by one or more processors, to the 3D segmented vessel tree geometric model at least one of a smoothing algorithm and a surface spline fitting algorithm. 
 
     
     
       8. The computer-implemented method of  claim 1 , generating the 3D segmented vessel tree geometric model by performing a back-projection the 2D segmented vessel images. 
     
     
       9. The computer-implemented method of  claim 1 , wherein the fluid dynamics machine learning model comprises at least one network of the type: convolutional neural network (CNN), autoencoder, or long short-term memory (LSTM). 
     
     
       10. The computer-implemented method of  claim 1 , wherein fluid dynamics machine learning model is a Navier Stokes informed deep learning framework configured to determine pressure data and velocity data over a 3D vessel space. 
     
     
       11. The computer-implemented method of  claim 10 , wherein the Navier Stokes informed deep learning framework comprises one or more methods of the type: Kalman Filtering, Physics-informed Neural Network, iterative assimilation algorithm based upon contrast arrival time at anatomical landmarks, and TIMI frame counting. 
     
     
       12. The computer-implemented method of  claim 1 , wherein assimilating the flow data over the sampling time period comprises determining pressure and flow velocity data for the one or more vessels over the sampling time period. 
     
     
       13. The computer-implemented method of  claim 1 , wherein assimilating the flow data over the sampling time period comprises determining pressure and flow velocity data for a plurality of connected vessels in the vessel inspection region. 
     
     
       14. The computer-implemented method of  claim 1 , wherein the computational fluid dynamics model comprises of one or more of:
 multi-scale 3D Navier-Stokes simulations with reduced-order (lumped parameter) models; reduced-order Navier-Stokes (1D) simulations with reduced-order models; or 
 reduced-order models for the segmented vessel tree geometric models. 
 
     
     
       15. The computer-implemented method of  claim 1 , wherein the lumped parameter boundary condition parameters are determined by the fluid dynamics machine learning model for one or more vessels in the vessel inspection region. 
     
     
       16. The computer-implemented method of  claim 15 , further comprising determining a lumped parameter model of flow for a first vessel and determining a lumped parameter model of flow for each vessel branching from the first vessel. 
     
     
       17. The computer-implemented method of  claim 10 , further comprising additionally applying the 3D segmented vessel tree geometric model to the fluid dynamics machine learning model. 
     
     
       18. The computer-implemented method of  claim 14 , wherein the reduced-models are graph-theory or neural network based reduced order models. 
     
     
       19. A computer-implemented method for assessing coronary artery disease, the method comprising:
 (a) receiving, by one or more processors, angiography image data of a vessel inspection region for a subject, wherein the angiography image data comprises angiography images captured over a sampling time period; 
 (b) applying, by the one or more processors, the angiography image data to a vessel segmentation machine learning model and generating, using the vessel segmentation machine learning model, two-dimensional (2D) segmented vessel images for the vessel inspection region; 
 (c) by the one or more processors, generating, from the 2D segmented vessel images, a one-dimensional (1D) segmented vessel tree geometric model of vessels within the vessel inspection region; 
 (d) applying, by the one or more processors, the angiography images to a fluid dynamics machine learning model and assimilating flow data for the vessel inspection region over the sampling time period; 
 (e) applying, by the one or more processors, the 1D segmented vessel tree geometric model and the assimilated flow data to a computational fluid dynamics model; and 
 (f) determining, by the one or more processors, a state of vessel occlusion for one or more of the vessels within the vessel inspection region. 
 
     
     
       20. The computer-implemented method of  claim 19 , wherein the computational fluid dynamics model is a graph-theory or neural network based reduced order model.